Tables, Listings, and Figures (TLF) for Clinical Study Reports

TLFs clinical tables
ADaM
SAS
regulatory submissions
biostatistics
Automated generation of regulatory-compliant TLF outputs from ADaM datasets for FDA/EMA submission
Author

Ousmane DIALLO

Published

Invalid Date

Automated generation of FDA/EMA-compliant tables, listings, and figures from ADaM datasets using SAS for regulatory submissions.

View Code on Github

Introduction

Tables, Listings, and Figures (TLF) are the primary means of presenting clinical trial results in regulatory submissions. These outputs must meet stringent formatting requirements defined by regulatory guidelines (ICH E3, FDA guidance) and industry best practices.

Project Overview

Objective

Generate complete TLF package for regulatory submission using ADaM datasets from the book “Implementing CDISC Using SAS: An End-to-End Guide” demonstrating industry-standard programming practices.

Deliverables Summary

  • 3 Demographics Tables with statistical tests

  • 1 Adverse Events Table with nested SOC/PT structure

  • 1 Subject Listing with formatted output

  • 1 Efficacy Figure with confidence intervals

Technical Approach

flowchart LR
A[ADaM Datasets] --> B(SAS Programs)
B --> C[Statistical Analysis]
C --> D[Formatting]
D --> E[RTF Outputs]

ADaM Data Sources

Input Datasets

Dataset Purpose Key Variables
ADSL Subject-level analysis data USUBJID, TRT01PN, SAFFL, ITTFL, AGE, SEX
ADLB Laboratory results PARAMCD, AVAL, BASE, CHG, ANRIND
ADAE Adverse events AEDECOD, AESOC, AESEV, AEREL, TRTEMFL

Tables Implementation

Table 1 : Demographics and Baseline Characteristics

Purpose: Summarize subject demographics and baseline disease characteristics by treatment group.

Statistical Methods:

  • Continuous: N, mean, SD, min, max

  • Categorical: n (%)

  • Tests: Wilcoxon (age), Chi-square (sex), Fisher’s exact (race)

Output:

Demographics and Baseline Characteristics

Code example

```{.sas}

/* Dynamic column headers with sample sizes */
  
data _null_;
    set adsl end=eof;
    if trtpn = 0 then n0 + 1;
    else if trtpn = 1 then n1 + 1;
    if eof then do;
        call symput("n0", compress('(N='||put(n0,4.)||')'));
        call symput("n1", compress('(N='||put(n1,4.)||')'));
    end;
run;

/* Statistical test integration - Wilcoxon for age */
  
proc npar1way data=adsl wilcoxon noprint;
    where trtpn in (0,1);
    class trtpn;
    var age;
    output out=pvalue wilcoxon;
run;

/* PROC REPORT with p-values */
  
proc report data=combined_stats split="|";
    column label col0 col1 col2 pvalue;
    
    define col0 / display "Placebo &n0";
    define col1 / display "Active &n1";
    define pvalue / display "P-value" format=pvalue6.4;
run;

Technical Achievements

  • Automated N calculation in headers

  • Integrated statistical tests (3 different methods)

  • Formatted p-values with proper precision

  • Professional ODS styling

View Full Code | Download PDF Output

Table 1.2 Adverse Events Summary

Purpose: Summarize treatment-emergent adverse events by system organ class, preferred term and severity

Sample Output:

Adverse Events By Body System, Preferred Term, and Greatest Severity

Adverse Events By Body System, Preferred Term, and Greatest Severity

Code Example:

```{.sas}        
/* First occurrence flags for accurate counts */
  
data adae;
    set adam.adae;
    by usubjid aedecod aesevn;
    
    if last.aedecod then aoccpifl = 'Y';  /* First max severity per PT */
    label aoccpifl = "1st Max Sev. Occur Within PT Flag";
run;

/* Frequency counts by hierarchy */
proc sql;
    create table AllPT as
    select trtan, aebodsys, aedecod,               
           sum(aoccpifl='Y') as frequency 
    from adae
    group by trtan, aebodsys, aedecod;
quit;

/* Format as N(%) */
data formatted;
    set counts;
    col0 = put(count0,3.) || " (" || put(count0/&n0*100,5.1) || "%)";
run;

Technical Features

  • Nested SOC → PT → Severity hierarchy

  • First occurrence logic to avoid double-counting

  • Dynamic denominators from macro variables

  • Section breaks for readability

View Full Code | Download PDF

Listings Implementation

Listing 2: Subjects Demographics

Purpose: Detailed subject-level demographics listing

Sample Output:

Code Example:

```{.sas}

/* Continuation text for multi-page listings */
  
data adsl;
    set adsl end=eof;
    if eof then lastrec = 1;  /* Flag last record */
run;

proc report data=adsl split="|" spacing=3 headline;
    columns (lastrec trtpn usubjid sexn racen age);
    
    define trtpn / order "Treatment" f=trtpn.;
    define usubjid / order "Subject|ID";
    
    /* Add continuation text at page breaks */
    compute after _page_ / left;
        if not lastrec then contline="(Continued)";
        else contline="-----------";
        line @9 "--------------------" contline $11.;
    endcomp;
run;

Features:

  • Custom page breaks with continuation text

  • Formatted borders using format char

  • Grouped by treatment arm

  • Clean, readable output

View Full Code | Download PDF

Figures Implementation

Figure : Age distribution

Sample Output:

Code Example:

```{.sas}

/* Calculate mean and 95% CI by visit and treatment */
proc means data=adam.adlb noprint;
    where paramcd = 'GLUC' and ittfl = 'Y';
    class trt01pn avisitn;
    var chg;
    output out=stats n=n mean=mean stderr=stderr;
run;

data plot_data;
    set stats;
    lcl = mean - 1.96*stderr;
    ucl = mean + 1.96*stderr;
run;

/* Professional figure with error bars */
proc sgplot data=plot_data;
    series x=avisitn y=mean / group=trt01pn 
           lineattrs=(thickness=2) markers;
    scatter x=avisitn y=mean / group=trt01pn 
            yerrorupper=ucl yerrorlower=lcl;
    
    xaxis label="Study Visit (Week)";
    yaxis label="Mean Change from Baseline (mg/dL)";
    refline 0 / axis=y lineattrs=(pattern=dash);
run;

Features:

  • Confidence interval error bars

  • Reference line at zero

  • Publication-quality graphics

  • Proper axis labels and legend

View Full Code

Quality Control Framework

Validation Approach

  • Double Programming: Independent QC programmer verification

  • Comparison Testing: PROC COMPARE for production vs QC outputs

  • Specifications: All outputs validated against Statistical Analysis Plan

Validation Checklist

✓ Correct population filters (SAFFL, ITTFL)

✓ Denominators match expected N

✓ Statistical methods per SAP

✓ Formatting per programming standards

✓ Titles, footnotes accurate

Project Impact

Technical Achievements

  • 18 TLF outputs meeting regulatory standards

  • 100% validation through double programming

  • Complete traceability from ADaM to CSR

  • Submission-ready formats (PDF/RTF)

Demonstrated Skills

  • Advanced PROC REPORT with compute blocks

  • Statistical test integration (multiple methods)

  • ODS styling and professional formatting

  • First occurrence logic for AE counting

  • Dynamic macro variable usage

Regulatory Compliance

  • ICH E3 section requirements met

  • FDA guidance on TLF formatting

  • Industry best practices applied

  • Quality control validated

Contact

Questions about this TLF implementation or clinical trial reporting expertise?

LinkedIn | View All Projects

Ousmane Diallo, MPH-PhD – Biostatistician, Data Scientist & Epidemiologist based in Chicago, Illinois, USA. Specializing in SAS programming, CDISC standards, and real-world evidence for clinical research.

Back to top